Data Sources & Research Standards

Our Research Philosophy

We follow a simple principle:

Use the strongest available source, verify important claims, acknowledge uncertainty, and never present estimates as official facts.

Not every CLAT-related data point is available in one official database.

For example, the Consortium may publish results, counselling lists, seat matrices, notices, and rank-based admission information, but a complete official marks-to-rank dataset for every candidate may not always be publicly available.

This means responsible analysis sometimes requires combining multiple reliable data points.

When we do that, we aim to be transparent about it.

Our Source Hierarchy

We do not treat every website, screenshot, social-media post, or data table as equally reliable.

We generally prioritise sources in the following order.

1. Official Consortium Sources

Where available, the Consortium of National Law Universities is our first reference point for CLAT-related information.

This may include:

• exam notifications;
• exam dates;
• application information;
• answer-key notices;
• result announcements;
• counselling schedules;
• admission instructions;
• official notices;
• other CLAT-related documents.

When an official source conflicts with an older third-party report, we prioritise the official information.

2. Official NLU Sources

For university-specific information, we may use official National Law University websites and admission documents.

This may include:

• seat matrices;
• category-wise seat distribution;
• domicile reservations;
• gender reservations;
• programme details;
• admission notifications;
• closing-rank lists;
• counselling updates.

University-specific admission policies can differ significantly, so we try to rely on the relevant NLU's own information wherever possible.

3. Official Counselling and Admission Data

Counselling data is particularly important when analysing NLU admission chances.

We may use:

• first allotment lists;
• subsequent counselling lists;
• vacancy lists;
• closing ranks;
• category-wise allotment data;
• upgrade movement;
• withdrawal-related changes.

These help us understand where admission actually ended rather than relying only on expected cutoffs.

4. Verified Candidate Data

Some marks-vs-rank relationships can only be reconstructed by comparing individual scores with actual ranks.

Where appropriate, we may review:

• publicly shared scorecards;
• verified candidate results;
• rank-and-score reports;
• voluntarily submitted result information.

We do not automatically treat every screenshot or claimed score as reliable.

Where possible, candidate-level data is cross-checked against broader patterns before being used for analysis.

5. Reputable Secondary Sources

When primary information is incomplete, we may use established education, legal-education, or examination publications.

Secondary sources may help with:

• historical score ranges;
• reported cutoff trends;
• exam analysis;
• marks-vs-rank estimates;
• counselling summaries.

We do not treat a secondary source as official merely because it ranks highly on Google or appears on a large website.

Its information is evaluated against other available evidence.

How We Handle Historical Marks vs Rank Data

Marks-vs-rank analysis requires particular care.

The relationship between score and AIR changes every year.

A score of 90 in one year may produce a substantially different rank in another.

To understand historical patterns, we may combine:

• reported candidate scores;
• actual AIRs;
• cutoff information;
• counselling ranks;
• top-rank score information;
• broader score-band trends.

Where a complete official score-to-rank dataset does not exist, the resulting table may represent an approximate historical range rather than an official mapping.

We aim to label such tables accordingly.

Why We Focus Primarily on the Latest Three Years

For rank prediction, recent data is generally more relevant than much older CLAT cycles.

CLAT evolves over time.

Changes may occur in:

• examination difficulty;
• question quality;
• score distribution;
• number of applicants;
• seat availability;
• admission rules;
• counselling behaviour;
• reservation structures.

Our rank-prediction methodology therefore gives greater importance to the most recent three examination cycles.

Older data may still be used for broader context, but it is not necessarily weighted equally.

How We Evaluate Data Quality

Before using a dataset or figure, we may consider questions such as:

• Is the source official?
• Is the information current?
• Can the data be independently verified?
• Does it align with other reliable sources?
• Is the number final or provisional?
• Is it specific to a category or programme?
• Does it refer to the first counselling list or the final closing rank?
• Has the source explained its methodology?
• Could the data have been affected by an answer-key revision?

Context matters as much as the number itself.

How We Handle Conflicting Data

Different sources sometimes report different values.

This can happen because of:

• different counselling rounds;
• revised answer keys;
• category differences;
• programme differences;
• outdated reports;
• rounding;
• incomplete candidate data.

When sources conflict, we may:

• prioritise official data;
• compare multiple reliable sources;
• identify whether the figures refer to different contexts;
• use a range rather than a single number;
• label the information as provisional;
• delay publication until the discrepancy is better understood.

We prefer uncertainty over false precision.

How We Research NLU Cutoffs

A cutoff can mean different things depending on context.

It may refer to:

• expected cutoff score;
• opening rank;
• first-round closing rank;
• final closing rank;
• category cutoff;
• domicile cutoff.

We try to identify exactly which type of cutoff is being discussed.

When comparing NLUs, we aim to avoid mixing unlike data.

For example, a first-list General closing rank should not be directly compared with a final-round domicile closing rank without explanation.

How We Research Seat Matrices

Seat matrices can materially affect admission probability.

We may review:

• total seats;
• General seats;
• category seats;
• domicile seats;
• women or horizontal reservation;
• special categories;
• programme-specific allocation.

Where an NLU updates its seat matrix, we aim to use the latest available version.

How We Assess Paper Difficulty

Paper difficulty is partly qualitative and partly visible through score distribution.

Our analysis may consider:

• section-level difficulty;
• length of passages;
• calculation intensity;
• ambiguity;
• number of high scorers;
• candidate performance trends;
• expert reviews;
• post-exam score patterns.

We avoid treating one person's opinion of difficulty as sufficient evidence.

How We Use Community Data

Student communities can provide valuable early signals immediately after the examination.

However, community data also carries risk.

Anonymous claims may be inaccurate.

Screenshots may be incomplete.

Scores may be self-reported incorrectly.

For this reason, Telegram posts, social-media comments, forum discussions, and anonymous score claims are not treated as authoritative on their own.

They may be used as early indicators only when supported by broader evidence.

How We Use AI in Research

AI-assisted systems may help us:

• organise large datasets;
• identify patterns;
• compare score bands;
• summarise information;
• support model-based analysis.

AI does not automatically make a source reliable.

Important data points still need to be assessed against underlying sources.

Our principle remains:

Source quality first. Analysis second. AI assistance third.

Research Standards for Predictions

A prediction is not judged only by how precise it looks.

We consider a good prediction one that:

• uses recent data;
• reflects uncertainty;
• adapts to paper difficulty;
• provides a realistic range;
• separates AIR from admission probability;
• accounts for category and domicile where relevant;
• can be updated as new information appears.

This is why we avoid artificial precision when the evidence does not justify it.

Our Approach to Numbers

We do not want numbers to look more precise than they really are.

When the evidence supports only a range, we use a range.

When a value is approximate, we indicate that.

When information is provisional, we say so.

A number such as AIR 900–1,100 can sometimes be more truthful than AIR 973.

Original Research and Internal Analysis

Some reports on CLATRankPredictor.com may contain analysis created by our own editorial or research team.

This may include:

• three-year trend comparisons;
• score-band studies;
• NLU probability models;
• rank volatility analysis;
• historical counselling analysis.

Where we publish original analysis, we aim to explain the methodology sufficiently for readers to understand how the conclusion was reached.

Updating Our Data

CLAT-related data is not static.

We may update our datasets when:

• new results are declared;
• answer keys change;
• counselling progresses;
• seat matrices are revised;
• new verified scorecards become available;
• official notifications are issued.

Where appropriate, pages may display a Last Updated date.

Data Corrections

If you identify a potential error in our data, we encourage you to report it.

Helpful correction requests include:

• the affected URL;
• the disputed figure;
• the correct figure;
• the relevant year/category/programme;
• an official or reliable source.

You can read our Corrections Policy for more information.

What We Do Not Consider Reliable Research

We do not intentionally base major conclusions solely on:

• anonymous social-media claims;
• unsourced screenshots;
• copied competitor tables;
• outdated articles;
• unverified WhatsApp or Telegram messages;
• unsupported AI-generated numbers;
• speculative claims presented as facts.

If evidence is weak, we would rather say that the information remains uncertain.

Our Research Standard

Before using an important data point, we aim to ask:

• Where did this number come from?
• Can we verify it?
• Is it current?
• Does its context match the conclusion we are drawing?
• Are we presenting it with the right level of certainty?

That is the standard we aim to apply across CLATRankPredictor.com.